Roland Orre's press text 040414 at 10:00
I'm Roland Orre, chairman and research director for NeuroLogic Sweden AB. A research and development company, developing multi relational data mining methods and artificial intelligence data analysis methods, based upon data warehousing principles. Since several years we have been developing early warning systems, and unsupervised pattern recognition methods for WHO, the World Health Organization, which are now becoming a standard within the pharmacovigilance sector.
Before this I was working at Asea Brown Boweri with human computer interaction where I was also developing expert systems to analyse power distribution failures, from which time I also have an EPO patent. This patent was a feature space approach to store analogue disturbances in a knowledge data base. An idea I got from my, at that time, recent experiences with neural networks, which I later did my PhD in.
Our company's view is that the patent system of today has serious flaws. One of these flaws is that software, which in reality is mathematics, interpretable by the abstract concept of a Turing machine, has become seen as a technology, which it is not, and as we have seen from United States almost everything has become patentable. As I said, software for me is mathematics and my favourite language, scheme, is very close to a mathematical theory called lambda calculus.
One of the original intentions with the patent system was that patents should be incentives to innovations and thus stimulate technological development, as we have seen from the software world, patents there have exactly the opposite effect. They slow down development, they are ineffient, as they can in reality only be thoroughly checked for free software, and they are trivial. In my Master thesis work I had for instance, used two trivial constructs which I ten years later got to know that these both constructs were patented at this time.
For a few years I was considering how these flaws of the patent system could be corrected, and I realized that a somwhat paradoxical way was to patent a method which could make inventions from peoples ideas, one of my old visionary dream inventions from the middle of the 80ies.
When I finally had understood, how to implemented this method, I patented it. The method is an artificial intelligence approach to do business, which is able to create inventions from peoples ideas, you could see it as collaborate innovation, in a somewhat similar way as the GPL, General Public License, works.
This patented method, which is a software solution, a business method and a system for manufacturing on demand, will, if this patent will be allowed within EU, be a cure for the patent system over time. It will speed up the technological development, it will correct the flaws of the patent system of today, as the system will produce new inventions it will also counteract trivial inventions. The system will enhance the patent system towards the real innovations. The patented method will also stimulate shaping of standards within knowledge representation and generic design. It will create win-win-win scenarios for the patent holder, for the customer and for the society as well, because it will also create new job incentives. One funny thing is, however, that if the patent system didn't have these flaws, that is, software and business methods being patentable, then I would hardly be able to patent this method.
Jörg F. Wittenberger
I'm Jörg F. Wittenberger, from softeyes.net; a free software developer since 1993. I've been working for Companies like Siemens, IBM and others. Mostly consulting in document, version and configuration flow control communication between my clients and their subcontractors and customers.
At a point I had the idea to use set theory to proof permission handling systems safe. (Which means to proof that no user can ever become impersonated by any other including administrative personnel.) I'm now actively marketing from a product based on that idea. But such a protection scheme only reaches as far as no malicious person can take physical action to modify sensitive data, let alone find a way to abuse administrative power.
To make my systems safe I had to assure the processes continue in the face of hardware failure or malicious attack from inside. To do so I reorganized the computation in such a way, that only the majority of all nodes of a network can define how the systems reacts over all. Now even the owner of a machine can not forge the global state of affairs. Equipped with my permission permission handling scheme, the system is now an intrusion resistant operating system, which serves undeniable processes -- "computation like power from the wall plug". I can grow now to a globally distributed operating system.
In the context of that network, which I'm calling "Askemos", several logical providers offer the execution of arbitrary operations as a service - following the model of power companies. All operations are completely defined by contract. Loyalty with respect to the contract is constantly checked during operation by comparison of cryptographic check sums. Furthermore in the abstract framework there is no concept of physical location or interaction - these are always done via the network.
I understand, that all objects and operations within Askemos are consequence of a business process and a priori without technical reference. Future theories could refine the definition of that criterion and proof other objects and operations as non-technical.
How relates the Askemos concept to patent law?
If program claims where allowed in patent law, how would that translate into the Askemos system?
Since the providers have a Carrier status with respect to the operation, they can't be liable for the combination of operations. This could be used as a patent law work around, which is obviously needed for the US but hopefully avoidable in Europe.
Program claims would furthermore conflict with freedom of speech and freedom of contract:
Askemos has a commercial counter part, where a business contract with different licensing conditions replaces the GPL. The system itself can work under any contract, as long as the contract provides "inalienable rights" and a Turing complete machine. Guaranties of some further requirements are just direct consequences. Related here is just one:
All devices used legally binding context must assure that they do not accidentally give up on rights of their users. Since it is easy to find that "security by obscurity" doesn't work, as all serious security experts will eventually confirm, devices must produce a proof that they follow a certain procedure. At that point we need to provide two things to our users a) the freedom to follow any code of their choosing and b) the chance to read the code beforehand.
If now article 3a of the parliamentarian decision where doped and "data processing" (which is to me the need of a "turing machine") became a field of technology - and thus due to TRIPS treaty subject to patent law - than anything in need of a computation became patentable.
Program claims are ill-logical since the support of a device instead of the brain to perform some mental operation doesn't make the operation itself a technical one. But the "media shift" to use computers for business matter makes contracts, trade and other business methods become subject to data processing. Our product just facilitates that trend. If we could not take non-patentability of turing machines for granted, computers would become useless in any legally binding context.